In the following the term “anastigmat” means an optical element or group of optical elements adapted to reduce astigmatism and/or aberrations including spherical aberration. See, e.g. Naumann/Schröder, Bauelemente der Optik, Carl Hauser Verlag München Wien, 6th ed., 1992, pp. 382-383 for a discussion of the term anastigmat. The term “Mangin mirror arrangement” means an optical device comprising a concave mirror and at least one negative powered lens proximal to the concave mirror wherein the concave mirror need not be in contact with the negative powered lens.
In the lithographic process for manufacturing semiconductor elements or the like, it is the usual practice to use a projection exposure apparatus for exposing a pattern image of a mask (or a reticle) onto a wafer (or a glass plate or the like) coated with photoresist via a projection optical system. Along with improvement of the degree of integration of semiconductor elements, the demand for achievable resolution of a projection optical system of the projection exposure apparatus is steadily increasing.
As a result, in order to satisfy the resolution requirement of the projection optical system, it is necessary to reduce the wavelength λ of the illuminating light (exposing light) and/or increase the numerical aperture NA of the projection optical system. More specifically, the resolution of a projection optical system is expressed by k·λ/NA (where k is a process coefficient). When assuming the refractive index of a medium (usually a gas such as air) between the projection optical system and the image field to be n, and the maximum incident angle to be θ, then, the numerical aperture NA on the image side can be expressed as n·sin θ.
Historically, resolution in microlithography has been improved either by increasing the numerical aperture (NA), or by reducing the wavelength of illumination light, or a combination of the two.
When it is tried to increase the numerical aperture by adopting a larger medium incident angle θ, the incident angle on the image plane and the outgoing angle from the projection optical system become larger, leading to an increase in reflection loss on the optical plane. It is impossible to ensure a large and effective numerical aperture on the image side. A technique is known for increasing the numerical aperture NA by filling an optical path between the projection optical system and the image field with a medium such as a liquid having a high refractive index. WO 99/49504 discloses a projection exposure method that irradiates exposure beams on a mask and transfers the pattern of said mask onto a substrate via a projection optical system, wherein when said substrate is moved along a predetermined direction, a predetermined liquid is passed along the direction of the motion of said substrate so as to fill the space between the end of the optical element on said substrate side of said projection optical system and the surface of said substrate, and discloses a projection exposure apparatus that irradiates exposure beams on a mask and transfers the pattern of said mask onto a substrate via a projection optical system, comprising a substrate stage that moves while holding said substrate, a liquid supply device that supplies a predetermined liquid along a predetermined direction via pipes for supply so as to fill the space between the end of the optical element of said substrate side of said projection optical system and the surface of said substrate, and a liquid recovery device that recovers said liquid from the surface of said substrate via said supply pipes and pipes for discharge arranged so as to sandwich the irradiation area of said exposure beams in said predetermined direction, and wherein when said substrate stage is driven to move said substrate along said predetermined direction, supply and recovery of said liquid is performed. The direction of the flow of the liquid may be changed according to the direction of the motion of the substrate. The projection exposure apparatus may be provided with a second pair of supply pipes and discharge pipes arranged at the location where said pair of supply pipes and discharge pipes would be if they were essentially rotated by 180°. The projection exposure apparatus may also comprise a liquid recovery device that recovers liquid supplied to between said projection optical system and said substrate.
U.S. Pat. No. 4,509,852 teaches using a photolithographic projection apparatus a mask having a pattern is imaged on a photosensitive layer coating a semiconductor substrate by a projection lens. To improve the resolving capability and to obviate adverse effects, e.g. standing waves and inhomogeneous exposure, the space between the substrate and the adjacent boundary face of a projection lens is filled during exposure with a transparent liquid having the same refractive index as the photosensitive layer.
However, a concrete proposal has not as yet been made regarding a configuration which ensures a large and effective image-side numerical aperture.
The theoretical resolution improvement of liquid-immersion is well known in microscopy, where oil-immersion dioptric objectives have for many years been designed with NAs greater than 1.0, but covering only a very small field of 0.5 mm or less. See, for example: “Modern Optical Engineering”, by Warren Smith, Third Edition, page 450, published by SPIE Press and McGraw Hill.
Liquid immersion applied to microlithography has also been proposed for many years, but has been slow to be adopted in production, no doubt because of practical difficulties. However, the theoretical advantages become stronger as “dry” projection lens NAs approach the theoretical limit of 1.0. These advantages have been described in, for example: “The k3 coefficient in nonparaxial λ/NA scaling equations for resolution, depth of focus, and immersion lithography” by Burn J. Lin published in JM3 1(1) 7-12 Apr. 2002.
More recent investigations into the practical issues of liquid immersion for lithography have also become more optimistic, for example: “Resolution enhancement of 157 nm lithography by liquid immersion” by M. Switkes and M. Rothschild, published in JM3 1(3) 225-228 October 2002. However, neither of these papers addresses the issues of optical design.
Early papers proposing liquid immersion lithography include: “Optical projection lithography using lenses with numerical apertures greater than unity” by H. Kawata, J. M. Carter, A. Yen and H. I. Smith, published in Microelectronic, Eng. 9, 31 (1989); “Fabrication of 0.2 μm fine patterns using optical projection lithography with an oil immersion lens” by H. Kawata, I. Matsumura, H. Yoshida and K. Murata, published in Japan, Journal of Applied Physics, Part 131, 4174˜1992; “⅛ μm optical lithography” by G. Owen, R. F. W. Pease, D. A. Markle, A. Grenville, R. L. Hsieh, R. von Bunau and N. I. Maluf, in the Journal of Vacuum Science Technology, B10-6, 3032˜1992; and “Immersion lithography at 157 nm” by M. Switkes and M. Rothschild, in the Journal of Vacuum Science Technology, B19-6, 2353-2001.
The recent Switkes paper is the most significant, in that it proposes the use of water as the immersion liquid for ArF (or KrF) laser light, perfluoropolyethers for F2 laser light, and starts to address the practical issues involved with a scanning wafer stage.
Another recent paper has started to address optical design issues for the relatively wide field of views used in lithography, partially disclosing liquid immersion dioptric microlithographic projection lens designs with NAs of greater than 1.0: “Development of dioptric projection lenses for DUV lithography” [4832-18] by Ulrich Wilhelm, Rostalski Hans-Juergen, Hudyma Russell M, published in SPIE Vol. 4832 IODC June 2002.
US 2001/0040722 A1 describes a catadioptric design which uses a V-fold mirror and two intermediate images. However, this is a small-field system (<1 mm), specifically intended for optical inspection, and there is no indication that the design could be applied to the much larger field sizes and extremely small residual aberrations and distortion required for microlithography.
“High numerical aperture lithographic imagery at the Brewster angle” by Timothy A. Brunner et al, in JM3 1(3) 188-196, October 2002, describes the fundamental disadvantages in terms of image quality, as the NA approaches 1.0 in a “dry” projection lens. These relate to vector imaging degradation that is made worse by Fresnel reflection losses at the resist interface, which more strongly reflects and loses the polarization orientation that would have given the better image quality inside the photoresist. This occurs most strongly at Brewster's angle, which corresponds to a NA of about 0.85.
We have investigated liquid immersion dioptric designs, and have found that for a NA of 1.0 and 26 mm field size the largest lens diameters need to be of the order of 330 mm, which is on the limit of available high quality fused silica, and beyond the limit for calcium fluoride. There is also a reduction in spectral bandwidth, in the same way that there is for “dry” dioptric lenses as the NA increases. A reduction in field size and an increase in reduction ratio above 4× would help these issues, but would make the “wet” lithography tools incompatible with current “dry” systems.
Known “dry” catadioptric designs have relatively small lens diameters and chromatic aberrations. However, they cannot be converted to liquid immersion only by adding a liquid to the space between the last element and the wafer. This would introduce a large amount of spherical aberration, which has to be compensated elsewhere in the design. Also, in simply adding a liquid, the NA does not increase, since the definition of NA already includes the refractive index.
Immersing the wafer in liquid is a necessary, but not sufficient, condition for being able to increase the NA up to the theoretical maximum equal to the liquid refractive index (˜1.4), rather than 1.0 in a “dry” system. For a constant magnification, paraxial geometrical optics theory (in particular, the Lagrange invariant) dictates that an increase of NA at the wafer has to be accompanied by a corresponding increase in NA all the way through the projection lens system. This results in an increase in lens diameters, and optical surface steepness, defined by the ratio D/R, where D is the clear aperture diameter and R is the radius of curvature. At the same time, chromatic and high-order aberrations increase rapidly with NA.
It is therefore not obvious to one skilled in the art of optical design that the NA of a “dry” projection lens can be increased in the ratio of the refractive index of the immersion liquid, without both an impractical increase in the lens size and complexity, as well as an unacceptable increase in residual aberrations.
Textbooks on optical design (e.g. Warren Smith, Modern Optical Engineering Third Edition, page 449-450, published by SPIE Press and McGraw Hill) describe the historical microscope immersion objective with a hyper-hemispherical convex surface (clear diameter/radius of curvature beyond hemispherical, where D/R=2) on the last element, opposite the plane surface in contact with the immersion liquid. Classically, this surface is designed to be either aplanatic, or close to the aplanatic condition. At the aplanatic condition there is zero spherical aberration, coma and astigmatism, and the marginal ray convergence angle is greater inside the lens element than before it by the ratio of the glass refractive index. Being close to this aplanatic condition minimizes spherical aberration and coma, and is a simple way of increasing NA, which is useful for a small field microscope objective, or systems such as the prior art US Patent Application US 2001/0040722.
For microlithography, which requires small aberrations over a much larger field size, such an aplanatic surface would give rise to higher-order aberration variations across the field, including oblique spherical aberration and coma. It is common practice to use, instead, a convex surface on this last element that is not at the aplanatic condition, but rather at or near the so-called concentric, or monocentric condition. In the concentric situation the marginal ray convergence angle inside the last element is identical to that incident upon it. Again there is zero spherical aberration and coma, but more importantly for a wide-field system there is zero sagittal oblique spherical aberration. See, for example, J. Dyson, JOSA, volume 49(7), p. 713 (July 1959), or, “Monocentric telescopes for microlithography” by C. G. Wynne, Optical Engineering, Vol. 26 No. 4, 1987.
J. G. Baker, The Catadioptric Refractor, The Astronomical Journal, Vol. 59, pp. 74-84 (1955) discusses pros and cons of a telescope which is based on a concept suggested by Schupmann (L. Schupmann, Die Medial-Fernrohre, Eine neue Konstruktion für groβe astronomische Instrumente, Teubner, Leipzig, 1899).